3.1. Silicon Pillars
Representative scanning electron micrographs of silicon pillars with three different cross-sectional geometries are displayed in
Figure 2a–c. They were fabricated on the same silicon wafers simultaneously and have an identical height of ~2.2 μm. C-shaped pillars with outer diameters of ~1.2 and ~2.7 μm are shown in
Figure 2a,b. Each C-shaped pillar consists of two sharp corners and curved edges as labeled in
Figure 2. Hollow-shaped columnar pillars were also examined and their micrographs are shown in
Figure 2c. The pillars were positioned in an orthogonal orientation. Physical dimensions of these structures are listed in
Table 1. Each of these pillar arrays covers a square area of ~3.5 mm × ~3.5 mm. The pillar's center-to-center distances of the three specimen groups—namely 1, 2, and 3—are approximately 6.2, 7.6 and 10.5 μm, respectively.
More importantly, the gaps between adjacent pillars—regions that allow portions of cells to extend into during the spreading process—are identical for all pillars (~5 μm). Badique et al. [
24] and Wang et al. [
8] have suggested that pillar spacing is an important parameter for nuclear deformation. Maintaining identical gap spacing between the pillars will reduce uncertainties of flow dynamic variations among pillars with different cross-sectional geometries and allow for a direct comparison of results from these three pillar groups. Unlike cylindrical pillars with axisymmetric geometry, the C-shaped pillars fabricated in this work provide a unique surface topography that produces non-axisymmetric mechanical stress states on the attached cells. Measured from the top-down views, the tip radii at the corners of the small and large C-shaped pillars are approximately 80 nm and 136 nm, respectively. Since the mechanical stress concentration factor increases with reduced tip radius [
28], the smaller C-shaped pillars with sharper corners and edges are expected to induce greater stress on the cells. In contrast, the 5.6 μm outer diameter hollow pillars are axisymmetric structures with low stress concentration and are provided as baselines for comparison with the C-shaped pillars.
The precise contact pressure profiles produced by these three pillar structures on the adherent cells are difficult to determine because they require detailed information on the pillar-cells’ three-dimensional contact profile, the mechanical properties of the live cells, and the directions and magnitudes of applied forces—which are hard to define as they change during the dynamic cell-spreading process. However, by using Equations (1)–(3) we can describe the effects of tip radius on contact pressure under simple loading geometries, such as spherical and cylindrical contacts on an elastic half-space. By reducing the spherical tip radius from 5.6 μm to 136 nm and 80 nm, the maximum contact pressure increases ~10.9 and ~16.0 times, respectively. The maximum contact pressures for smaller cylindrical contacts are ~5.4 and ~7.4 times greater, respectively. In both contact cases, the maximum pressure increases with the reduced radius of the structure. As mentioned above, it is important to note that the actual stress profiles experienced by the adherent cells can be more complex than the two simple models evaluated here; however, they do serve as a demonstration of the pillar corner and edge radius effects.
3.3. Cells on Patterned Silicon Substrates
Representative confocal micrographs of PC3 cells incubated on small (group 1) and large C-shaped (group 2) pillars for 72 h are shown in
Figure 3. An example of PC3 cells incubated on small C-shaped pillars (group 1) for 72 h is displayed in
Figure 3c,d. These micrographs were collected at a focal plane below the top of the pillar, as schematically illustrated in
Figure 3e. A composite micrograph of blue DAPI, red phalloidin, and gray-scale optical reflection images of the silicon-patterned structures is displayed in
Figure 3c. The inset schematic drawing indicates the orientation of the small C-shaped pillars on the substrate. Pillar locations are highlighted with solid arrows unless otherwise stated. This micrograph shows many filopodia and lamellipodia are observed on this cell, with the majority of these being attached to the substrate. It is interesting to note that some appear preferentially bonded to the pillars. Such selective attachment behaviors of actin structures on small pillars have been previously reported by Albuschies and Vogel [
37] with human dermal foreskin fibroblast cells. In addition, Jahed et al. [
38] showed 3T3 Swiss Albino fibroblasts cells preferentially sensing and remaining attached to metallic nanopillars.
Figure 3d shows the DAPI-stained areas of this cell in gray-scale. Detailed inspections of this image reveal the locations of C-shaped pillars as small solid dark spots (highlighted with white solid arrows). This micrograph also showed other features with a varying intensity of DAPI fluorescence which may represent the locations of different sub-nuclear organelles. At least eleven silicon pillars can be identified surrounded by the nucleus displayed in
Figure 3d with six located at the center of the stained region, while the rest are positioned along the edges.
Figure 3d shows this nucleus to have distinctly bulged edges (highlighted with yellow dashed arrows) at the top, bottom, and left portions of the nucleus. They are likely created by the flow of membrane-bound nuclear material in between the stiff silicon pillars during the cell spreading process. The nucleus and its membrane are expected to experience highly-localized mechanical stress at the pillar locations where the material flows are restricted and the nuclear membrane is severely deformed in order to conform to the shape of the silicon pillars.
Another important characteristic feature observed in some of the cells deposited on the small C-shaped pillars (group 1) are distinct fine dark line structures revealed in the DAPI-labeled micrographs collected at focal planes below the pillar top, as schematically illustrated in
Figure 3e.
Figure 3f shows examples of such fine dark lines in a DAPI-stained area radiating outward from the pillar locations and the inset drawing indicates the orientation of the small C-shaped pillars on the substrate. The locations of four pillars are highlighted with solid arrows. These dark lines, less than a micron wide, can be clearly seen emanating from all four pillars located near the center of the nucleus, as displayed in
Figure 3f. They are uniquely different from the randomly-distributed and irregularly-shaped dim background features produced by other sub-nuclear organelles. The lack of DAPI fluorescence signal in these fine line structures indicates a low concentration of nuclear DNA in these features. Some nuclei, as shown in
Figure 3g,h, contain line features that are not connected to adjacent pillars and have lengths in the micron scale. Additional micrographs are revealed in
Supplementary Figure S1 to demonstrate the reproducibility of these dark line features. The DAPI-labeled micrographs shown in
Figure 3f–h also reveal that the pillar's dark spots resembling filled semi-circles rather than their true C-shaped geometry. This indicates that the pillars are not in direct contact with the DNA. A possible reason is that the nuclear envelope and the plasma membrane remain intact at the cell-pillar interface and provide a gap between the DNA and the pillar. These pillars are unable to rupture the cell membranes and make direct contact with DNA materials. Instead, cells putatively engulf and wrap around the pillars. This is expected as several previous studies [
39,
40,
41,
42] have shown that micropillars are unlikely to penetrate the cell membranes due to their large dimensions and small height to diameter aspect ratios. The narrow dark spacing between the pillar and the DAPI stained DNA shown in
Figure 3f,h may be filled with other cellular components, such as the plasma membrane, cytoplasmic components, the nuclear envelope, or other sub-nuclear organelles. Hence, the dark spots do not resemble pillar true geometries.
Random inspection of 55 cells deposited on small C-shaped pillars (group 1) revealed that 35 cells (64 ± 6%) exhibited these dark line structures where the data spread corresponds to one standard error. These results are plotted in
Figure 4a. The length distributions of 132 dark lines surrounding small C-shaped silicon pillars were measured and displayed in
Figure 4b. This cumulative probability plot shows the line lengths are in the range of ~1.3 and ~7.3 μm, with the average value of 3.8 ± 1.4 μm. The data spread represents one standard deviation. On average there are ~3.8 lines formed in each cell that contains dark lines. Among the 132 lines inspected, 104 of them (79% of the line population) extended from the sharp corners, while 28 lines (21%) emanated from the curved edges. This demonstrates that the sharp corners of the C-shaped pillars are the more likely locations to form dark line structures.
Careful inspection of PC3 cells incubated for approximately 72 h on large C-shaped pillars (group 2) revealed similar sub-micron-scale dark line structures, as shown in
Figure 3i. The large C-shaped pillars surrounded by the nucleus are labeled with arrows and the separation distance between the pillars is identical to the small C-shaped pillars (~5 μm). This confocal micrograph shows that the dark spots where the pillars are located do not bear a resemblance to a true pillar C-shape but instead have filled-triangular profiles. Interestingly, several of the fine dark line features extend from the tips of these triangles—an indication that they may be structurally connected. Random inspections of 38 PC3 cells attached on large C-shaped pillars (group 2) showed that 17 of them (45 ± 8%) exhibit line features as shown in
Figure 4a. These results show that fewer PC3 cell nuclei (45% vs. 64%) exhibit line structures when they are deposited on the large C-shaped pillars in comparison to the smaller counterparts.
The length distributions of 27 dark line structures surrounding the large C-shaped pillars are plotted in
Figure 4b. The average dark line length in these cells is 4.2 ± 1.9 μm where the data spread corresponds to one standard deviation. These results are statistically indistinguishable from the small C-shaped pillars results. The average number of dark lines observed in cells contacting large C-shaped pillars is ~1.6—two times fewer than those observed in small C-shaped pillars of 3.8. One primary reason is a reduced number of dark lines originated from the curved edges of the large C-shaped pillars. Inspection of cells on large C-shaped pillars show 24 out of 27 lines (89%) originated from the two sharp corners of pillars but only three lines (11%) were extended from the curved edges. The lack of dark lines emanating from large C-shaped pillar curved edges may suggest that their formation may be related to the dimension of the pillars. Large C-shaped pillar diameters are more than two times larger than the small C-shaped counterpart (1.2 vs. 2.7 μm).
Results shown in
Figure 3 and
Figure 4 indicate that the dark line formation process depends on the pillar’s cross-sectional geometry and may be influenced by the sharp corners and curved edges of the small C-shaped pillars. These sharp features are expected to magnify the mechanical forces produced by the cytoskeletal structures pushing downward on the nuclei [
4,
28]; hence, greater indentation forces are applied to the cells. To test this hypothesis, cells were deposited on hollow-shaped pillars fabricated on the same substrate as the C-shaped pillars. The separation distance between the hollow pillars is identical to the C-shaped structures of ~5 μm in order to reduce the uncertainties related to dynamic flow of material between the pillars. Since these hollow pillars have axisymmetric geometry with smooth, curved surfaces, the amount of induced stress is expected to be low when compared to C-shaped pillars with sharp corners.
A representative image of a PC3 cell that was incubated on hollow pillars for approximately 72 h is shown in
Figure 3j and does not show distinct line structures. Random sampling of 28 cells indicated that only five nuclei (18 ± 7%) showed dark line structures, as shown in
Figure 4a. These results signify that the important factor for the dark line formation is pillar’s cross-sectional geometry. The line length distributions on these cells are shown in
Figure 4b and show the dark lines observed on hollow-shaped pillars have average lengths of 5.1 ± 2.6 μm. While the average length is longer than those observed in C-shaped pillars, the differences are statistically insignificant due to the large data standard deviation. Furthermore, the average number of lines observed surrounding each cell is 1.8 which is approximately the same as those on the large C-shaped specimen. Experiments were repeated with cells incubated for approximately 24 h and showed similar dark line structures. Examples of 24 h cell incubation on small C-shaped pillars (group 1) and large C-shaped pillars (group 2) are shown in
Supplementary Figure S2a,b respectively.
To understand the three-dimensional configurations of these line features, additional inspections of the PC3 cells were conducted at multiple focal planes. High-magnification DAPI-stained micrographs of the fine line structure from the large nucleus displayed in
Figure 5a were collected and are shown in
Figure 5b–e. The locations of these focal planes are schematically illustrated in the accompanied diagram. The z-section image sequence begins at a focal plane slightly above the top of small C-shaped pillars, and sequentially downward to the pillar base. White solid arrows in the micrographs indicate where the two pillars are located. No fine line structures or pillar dark spots are observed in the micrographs taken at focal planes above the pillar tops (
Figure 5b). Faint impressions of two C-shaped pillars appear in
Figure 5c with the focal plane located near the top of the pillars, but no dark line is observed in this micrograph. When the focal plane is located below the pillar top level, as the micrograph shown in
Figure 5d, a faint fine dark line that connects the pillars is clearly visible. In addition, the two pillar dark spots displayed in
Figure 5d show they do not resemble the true pillar C-shaped profile but instead a tear-drop shaped geometry where the tail portions of the feature narrows to form fine dark lines connecting adjacent pillars. This indicates that the pillar's dark spots and the fine dark lines are structurally connected.
Furthermore, the tear-drop shaped and fine dark line structures are observed at other focal planes further down toward the substrate (see
Figure 5e). This suggests that these observed features are cross-sectional views of a thin continuous vertical slit connecting adjacent pillars. It begins below the top of the pillars and extends downward. The geometric structure of this nanometer scale slit is further confirmed by the orthogonal views of this cell nucleus as shown in
Figure 5f where the cross-section locations are labeled with dashed lines. A nano-slit is clearly visible on the xz-plane.
Schematic cross-sectional drawings of nuclear DNA elements surrounding two C-shaped pillars are shown in
Figure 6. The drawings indicate that the DNA elements do not make direct contacts with the pillars. Instead, tear-drop shaped gaps are formed between the DNA and the pillars where they may be filled with the plasma membrane, nuclear envelope, or other sub-nuclear organelles that were not stained by DAPI. To the best of our knowledge, this is the first literature report of silicon micropillar-induced nano-slit structures within human prostate cancer cells. Confocal micrographs reveal that these continuous slits of materials are depleted of DNA.
The formations of these nanometer-scale thick slits are not only observed in cells deposited on a uniform array of C-shaped pillars, but have also been confirmed in cells that are simultaneously contacting pillars with different cross-sectional geometries, as revealed in
Figure 7a,b. These confocal fluorescence micrographs show the DAPI and composite images of a cell that is adhered concurrently to two different arrays of hollow silicon pillars (group 3) and small C-shaped (group 1) geometries. The majority of the cell nucleus is surrounding the small C-shaped pillars (highlighted with arrows) while the remaining part of the cell is extended to the hollow pillars. Even though the cell contacts drastically different surface topographies, the DAPI-strained DNA materials still show slits as those shown above. It is remarkable that slits can be observed in the two C-shaped pillars that are less than 5 μm away from the hollow pillar arrays (marked with dash arrows). However, it is unclear if the orientations and the lengths of these slits have been influenced by the nearby hollow pillar arrays. The evidence presented here indicates that the slit formation mechanism may be driven by a localized effect at the micron-scale that is determined by the pillar geometries.
The confocal micrographs presented above show that sub-micron-scale stiff silicon pillar structures with C-shaped geometries produce nanometer-scale slit features in the PC3 cells. Results also revealed that not only can the nuclei be reshaped, but they may also be sensitive to the cross-sectional geometries of the surface topography. The exact origin and mechanisms of these slit formations have yet to be identified. One possible explanation may be related to the pillar/cell contact geometries. As cells spread on textured silicon surfaces their cytoskeletons push the nuclei and other sub-cellular organelles downward and press against the pillars [
22,
23]. This resembles nano-indentations of viscoelastic materials (cytoplasm and nuclei) contained within elastic membranes (plasma membranes and nuclear membranes) by a series of complex-shaped silicon cylinders. Hanson et al. showed that nuclear deformation on nanopillars is determined by nuclear stiffness as well as cytoskeletal forces, and that the geometry of nanopillar arrays highly influences nuclear deformation. As live cells press against the silicon pillars, the plasma membrane, cytoplasm, and nuclear envelope deform and occupy the space in between the pillars. For asymmetric hollow pillars, deformations near the contact rims were evenly distributed. However, membrane deformations on complex-shaped pillars, such as C-shaped structures, were more severe near the two sharp corners. This may cause the plasma membrane and nuclear envelope to fold inward and form creases near the corners. Since the membranes do not contain DNA, the creases formed resemble thin dark lines or slits, as illustrated in
Figure 3 and
Figure 5.
This work demonstrates that PC3 cell mechanical contact responses to stiff silicon micro-pillars are more complex than previously understood and open a new direction of investigation in this field. Future research should include in situ live cell time-lapse microscopy imaging studies of these cancer cells with different pillar geometries. This focus could gain temporal information about how the slits are formed and help develop an understanding of whether the presence of the nano-slits will prevent or restrict the transport of the nuclear material between different parts of the nucleus. The effects of these structures on basic cell biological functions, such as proliferation, metabolism, mitosis, and cell division, should be investigated and compared between cancer and normal cells. During typical cell division processes, DNA molecules are replicated and chromatin molecules are being condensed into chromosomes (prophase). Eventually, the chromosomes or sister chromatids are being pulled by the microtubules to the opposite ends of the cells (metaphase and anaphase). However, it is unclear how the slits, which may act as physical barriers, affect the chromosome replication processes and microtubule motions. Even if cells were successfully divided, it is unclear if the presence of slits will induce chromosome replication errors. To address these questions, further experiments are needed to compare the DNA sequences among cells grown on the patterned surfaces with different passages. Other essential macro molecules, such as ribonucleic acids (RNAs) and peptide chains, should also be analyzed from different passages to understand if the cell functions are compromised.
Additional investigation is also required to determine how the response of PC3 cells to silicon micropillars compares with other cancer and healthy cell types. One type of the potentially healthy cells to be investigated are human fibroblast cells. Their average nuclear dimensions are similar to the PC3 cells, on the order of tens of microns, when they adhered and spread on the patterned silicon surfaces. This type of study will provide a comparison of surface topographic responses on normal and cancerous cells. Investigations of slit formations should also be conducted on other immortalized cancerous cell lines, such as HeLa cervical cancer cells, to determine if the slit formation is a mechanical contact phenomenon unique to cancerous cells or PC3 cells. Finally, other pillar cross-sectional geometries with higher stress concentration points should also be investigated to understand the effect of introducing a stress field to sub-nuclear organelles.